Off-Grid Monitoring: Power Budgets for Sites With No Mains

Last updated: 28 August 2026

In short: Off-grid monitoring almost never fails because of radio range. It fails in February, because the solar array was sized on an annual average instead of a British December. A gateway drawing 3.6 W needs roughly 150 W of panel and around 50 Ah of lithium storage to run through midwinter. That ratio surprises nearly everyone.

The site is a reservoir compound, a rural pumping station, a quarry weighbridge or a stretch of embankment. There is no mains supply and no prospect of one, so the design becomes solar, battery and a radio link.

The radio part of off-grid monitoring is the easy part. This guide is about the part that actually strands sites, which is the power budget, and specifically what happens to it between November and February.

Why does off-grid monitoring fail in winter?

Because UK solar yield collapses far more than most people assume, and it collapses exactly when you most want the data. Modelled figures from the European Commission’s PVGIS database, averaged across UK postcodes, put December output at 29 kWh per kWp against 119 kWh per kWp in May.

That is a four-fold swing. A system designed on the annual mean is designed for a month that does not exist, and it will run out of charge somewhere around the second week of a dull January.

The cruel part is the correlation. Storms, floods, freezing pipes and access problems all cluster in the months when your off-grid monitoring has the least energy to report them with.

Work out the load first, in watt hours per day

Every off-grid monitoring design follows from one number, and it is not the panel size. It is the daily energy consumption of everything at the site, in watt hours.

Take a LoRaWAN gateway as the anchor load. The Milesight UG67 datasheet quotes power consumption of typically 3.6 W with a maximum of 4.8 W, powered either by 802.3af PoE or 6 to 12 V DC through an M12 connector.

At 3.6 W continuous that is 86.4 watt hours per day. Use the typical figure for sizing the panel and the maximum figure for sizing the cable and the fuse, because those two jobs have different failure modes.

Now the point that catches people. A LoRaWAN gateway cannot be duty cycled. Battery sensors sleep between readings, which is why they last for years, but the gateway has to listen continuously or it misses uplinks that were only ever sent once. Your anchor load runs 24 hours a day, every day, and no clever firmware changes that.

Add the backhaul. If the site uses 4G rather than Ethernet, that draw is inside the same envelope on an integrated unit, but a separate router is a second continuous load to count.

Sizing the array for the worst month, not the average one

Take the December figure of 29 kWh per kWp per month. Divide by 31 days and you get roughly 0.94 watt hours per day for every watt of panel you install.

To cover 86.4 watt hours a day you therefore need about 92 W of panel, and that is at perfect efficiency with nothing lost anywhere. Real systems do not work like that.

Derate for reality. A charge controller takes a few per cent, battery round trip efficiency takes ten to fifteen, and cable losses, module soiling, imperfect tilt and low winter temperatures take more. A combined figure of around 70 to 75% is a defensible planning assumption.

That pushes 92 W up to roughly 125 W, and once you add any margin at all you are specifying a 150 W panel to run a 3.6 W device. The ratio is around forty to one, and it is the single most useful thing to understand about off-grid monitoring before you price a site.

Sizing the battery for a run of dull days

In off-grid monitoring the panel handles the average day. The battery handles the week when there is no useful sun at all, which in a British winter is entirely normal.

Five days of autonomy at 86.4 watt hours per day is 432 watt hours of usable energy. Usable is the operative word, because you never get the nameplate capacity.

• Lithium iron phosphate at a sensible 80% depth of discharge needs about 540 Wh of nameplate, which is roughly 45 Ah at 12 V.
• Sealed lead acid at 50% depth of discharge needs about 864 Wh, which is roughly 72 Ah at 12 V, and it is far heavier.
• Cold derating then applies on top. Lead acid loses a meaningful fraction of its capacity near freezing, so a nominal 72 Ah becomes closer to 100 Ah in a specification you can trust in January.

Lithium iron phosphate is usually the right answer for off-grid monitoring on cost per usable cycle, but it has one condition worth knowing: most cells must not be charged below 0°C without a heater or a low temperature cutoff in the battery management system. Check that the battery you buy has one.

A worked off-grid monitoring example

Put the numbers together for a real shape of site: a rural pumping station with a gateway, six battery sensors and a 4G backhaul, with no mains and no realistic prospect of a supply.

Load. The gateway is the only continuous consumer at 3.6 W, because the sensors run on their own primary cells and contribute nothing to the shared budget. Daily energy is 86.4 Wh.

Array. December yield of 0.94 Wh per watt of panel per day means 92 W before losses, about 125 W after a 70 to 75% system efficiency, and 150 W once you allow any margin at all.

Storage. Five days of autonomy is 432 Wh usable, which is a 45 Ah lithium iron phosphate battery at 12 V with an 80% depth of discharge and a low temperature charge cutoff.

What that costs you. The solar and battery hardware on an off-grid monitoring site of this size is routinely a larger line item than the gateway it exists to power. That is the correct outcome, not a sign you have specified it wrong.

The instructive part is the ratio. A 150 W array and a 45 Ah battery to run one 3.6 W device looks absurd on a spreadsheet, right up until the first off-grid monitoring site you built to a summer budget goes silent in the second week of January.

Mounting, tilt and the things that halve your yield

A steep winter tilt is the cheapest off-grid monitoring upgrade available. Angling the panel closer to vertical, around 50 to 60 degrees from horizontal, catches more of a low winter sun and sheds snow and leaf litter instead of collecting them.

You will lose some summer yield doing this. That is the correct trade, because your system is not constrained in June.

Shading is the other yield killer, and it behaves worse than people expect. Partial shade on part of a panel can cut output disproportionately, not proportionally. Survey the site for winter shadows specifically, because a tree line or building that clears the panel in July will cross it at midday in December when the sun is low.

Face the array as close to due south as the mounting allows, and keep it reachable. An array that cannot be cleaned without a cherry picker will not be cleaned.

Making the site tell you it is in trouble

The worst failure mode in off-grid monitoring is silence, because silence is ambiguous. A site that has gone quiet might be flat, might be flooded, or might have a failed antenna, and each needs a different van with different parts in it.

Design off-grid monitoring for a dying gasp. The UG67 includes a capacitor specifically to send alarms in the event of power failure, so the site can announce its own death rather than simply disappearing.

Report battery voltage as an off-grid monitoring channel in its own right, alongside whatever the site is actually there to measure. A slow decline in daily minimum voltage across three weeks is a site telling you it will fail next month, which is a maintenance visit you can plan rather than an emergency you cannot.

Set the alert on the trend, not only on the threshold. By the time voltage crosses a low cutoff you have hours, not weeks.

When solar is the wrong answer

Sometimes solar is the wrong basis for off-grid monitoring, and saying so early is cheaper than proving it in February.

If the site is deeply shaded, north facing, inside a structure, or subject to vandalism, a large array is a liability rather than a solution. Primary lithium battery packs sized for a two or three year replacement cycle can be a better economic answer for a low duty load, particularly where the sensor can report daily rather than continuously.

If there is no cellular coverage either, backhaul becomes the constraint before power does, and satellite links carry their own significant power and cost implications that need to be designed in from the start rather than added later.

Be honest about access as well. A site you can only reach with a boat, a key from a third party, or a two hour drive should be specified with more autonomy than the maths strictly requires, because the cost of a visit dominates the cost of a bigger battery.

Frequently asked questions

How big a solar panel does a LoRaWAN gateway need in the UK?

For a gateway drawing 3.6 W continuously, plan on roughly 150 W of panel to survive December, once system losses and margin are included. A 50 W panel will work beautifully from April to September and strand the site in winter.

Can I duty cycle the gateway to save power?

No. Sensors sleep, gateways cannot. A LoRaWAN gateway that is asleep misses uplinks permanently, because Class A devices transmit when they choose and do not retry indefinitely.

Lithium or lead acid for off-grid monitoring?

Lithium iron phosphate suits off-grid monitoring in most cases, for usable depth of discharge, weight and cycle life. Confirm the battery management system has a low temperature charge cutoff, because charging most lithium chemistries below freezing damages them.

How many days of autonomy should off-grid monitoring have?

Five days is a reasonable UK baseline. Increase it for sites that are hard to reach, and for anything where the data is used for a safety or compliance duty rather than for convenience.

Does a bigger battery fix a small off-grid monitoring panel?

Only temporarily. In off-grid monitoring the battery covers gaps between charging days, but if the panel cannot replace the daily draw across a whole month the system runs down regardless of storage. Size the panel for December first, then the battery for the dull run.

The short version

Off-grid monitoring is an energy problem wearing a radio costume. Work out watt hours per day, size the array against the December figure rather than the annual mean, give it five days of usable storage, tilt it steeply, and make the site report its own battery voltage so it can warn you before it goes dark.

For the connectivity layer, our guide to what LoRaWAN is and where it fits covers range and battery behaviour, and the LoRaWAN starter kit guide covers what a first deployment actually needs. For remote assets specifically, tank level monitoring is the most common off-grid application we quote.

The Milesight UG67 outdoor gateway is priced and in stock. Add it to a quote and we will come back within one working day.